Fuzzy Application Library/Technical Applications/Nuclear Fusion

First Fuzzy Logic control of nuclear fusion experiment

By Bill Edisbury, Npower Ltd. UK, May 1999

A solution to the problem of how to achieve controlled nuclear fusion may be nearer than we think if plans to use fuzzy logic to control the key features of the fusion process get the go-ahead at the UKAEA Culham Science Centre in Oxfordshire. Early trials of fuzzyTECH - fuzzy logic software supplied by Manchester- based Npower - have allowed UKAEA physicists to monitor the fuel density in their research machines far more accurately than has been possible previously. This will be the first time that fuzzy logic software has been demonstrated in a nuclear fusion context. Until now physicists have grappled with the problem of controlling the plasma (ionised gas) used in nuclear fusion research using fast computers but even these have difficulty keeping up with controlling each part of the complex fusion processes simultaneously. Fuzzy logic overcomes this by taking a holistic approach to the fusion process, enabling physicists to concentrate on controlling just a few key variables and leaving the fuzzy logic software to 'mind' the rest.

"Improving control of the plasma will be a great step forward in the realisation of power from nuclear fusion," says Tom Todd, manager, Fusion and Industry, UKAEA Culham Division. At present, the plasma pulse can last from a few milliseconds up to a minute depending on the size of the fusion vessel: the larger the vessel, the longer the fusion process lasts (although other factors play a role). "Our eventual aim is to achieve a process that is stable for months at a time without the need for a huge machine and fuzzy logic may provide part of the solution," says Todd. So far fuzzy logic has been piloted in monitoring the density of the fusion plasma, although there is growing interest in extending its application to control the plasma shape. If plans to use fuzzy logic for feedback control go ahead, it is hoped that this will extend the stable operation of the plasma, enabling physicists to conduct in-depth research into the optimum plasma shape more thoroughly. Currently this can be either like a 'beach ball' or ring-doughnut, but other shapes are also possible, the trick is to find which is best. Nuclear Fusion Nuclear fusion works by taking heavy isotopes of hydrogen (deuterium and tritium) and exposing them to high pressure and temperatures (100 - 200 million degrees centigrade, far in excess of the temperature of the core of the sun which is a mere 2 million degrees centigrade). The fusion reaction produces a helium nucleus, known as an alpha particle, a high- energy neutron and, at the same time, releases energy.

The high temperature of the fusion process forces the fuel to be in the form of plasma, when the atoms are free of their electrons. Strong magnetic fields are used to confine the plasma, preventing it from touching, and melting the walls of the fusion vessel. The energy released by nuclear fusion would be transferred via cooling circuits to heat-exchanging boilers, with electricity generated in steam turbines in the conventional way. The eventual commercialisation of nuclear fusion will depend upon the ability to produce a stable plasma, one that produces self-sustaining energy over an extended period. Cracking the stability problem relies upon being able to hold the plasma accurately to a specific shape and in place using magnetic confinement. Controlling the shape and density of the plasma are key to managing the stability of the plasma: fuzzy logic offers the potential to solve both problems. Fuzzy logic for enhanced feedback control Variations in the refractive index of the plasma with its density mean that an interferometer, typically using infrared laser beams, can measure the density by counting the number of wavelengths (fringes) produced in passing through the plasma. Until now 'traditional' fringe counters have been unable to cope with sudden changes in density, producing erratic jumps in the signals which make real-time control impossible. Fuzzy logic has been shown to 'smooth out' the fringe readings, enabling real-time control of the plasma density.

Measuring the shape of the plasma involves using an array of magnetic probes, and other diagnostic instruments, to monitor the edge of the plasma. The output of the diagnostics can be processed and incorporated into real-time feedback control by fuzzy logic software, enabling physicists to concentrate on just a few key variables to control the fusion process. A key feature of the fuzzy logic controller would be the ability to automatically and smoothly adjust the priorities of a number of controlled variables as each plasma "shot" progressed, permitting more variables to be controlled than there are control parameters available to the operators. - ends - Editor's Note Npower Based in Manchester, Npower offers both software and consultancy to exploit the power of business intelligence tools such as fuzzy logic or neural networks, case based reasoning and knowledge management. Npower's customers include Abbey National, UBS, Barclays, the Dental Practice Board, Cranfield, British Steel, UKAEA and Zeneca. DTI Smart Software for Decision Makers Npower is a technology provider to the Department of Trade and Industry's Smart Software for Decision Makers (SSDM) Programme. This three-year initiative aims to help UK businesses become more aware of the commercial benefits of adopting emerging intelligent software technologies for converting raw data into business intelligence. Programme material available to businesses ranges from comprehensive "how to" guides including business and technology briefs, to a getting started guide, case studies and directories of product and service suppliers, as well as user clubs that assist members in developing demonstrator projects.